In three essays, I show the importance of model choice in distinguishing the expected from the unexpected. For event studies, I show that presence of event endogeneity is sensitive to the choice of benchmark to measure normal returns. For the case of correlated earnings announcements, I show that correctly modeling the heterogeneity in expected returns explains apparent overreaction in a firm’s stock price to its competitor’s earnings announcements. For equity premium puzzle, I show that allowing for the presence of a temptation effect helps explain the surprisingly high historical returns to U.S. stocks. |